Introduction

The East Asian Social Survey (EASS) is a biennially study conducted in four East Asian countries of China, Japan, Republic of Korea, and Taiwan. The 2006 survey was conducted to study East Asian family, 2008 for globalization and culture, 2010 for health, 2012 for social capital, 2014 for work life, and 2016 again for family. The data collection for the 2018 module has also started this year. I chose to work with the 2012 social capital module to explore the differences and similarities among these four countries so often grouped together in Western academic wings of social and behavioral sciences, and challenge the predominant one-dimensional view that groups these countries as one singular entity by discovering nuanced differences among the four vastly different countries.

Grouping the variables

There is adequate explanation provided for each of these variables are surveying in the codebook (refer to pp. 155-167 or pp. 101-147). To help our analysis, I first went through the codebook to identify which of these variables capture demographic information for each of the survey respondents which we may be able to single out as independent variables. From the rest, I grouped the related items together by theme to systematically process appropriate analyses. From the list/groups, I picked and chose the characteristics I was interested in exploring, comparing, and contrasting with one another.

Potential independent variables (demographic indicators)

  • V2: Country/Region
  • V13: Organizations Participated Most Actively in the Last 12 Months
  • SEX: Sex
  • AGE: Age
  • MARITAL: Marital status
  • MAIN: Main status
  • RELIG: Religious main groups
  • TOPBOT: Top bottom self-placement
  • URBRURAL: Self-assessment of community type
  • CN/JP/KR/TW_REG: region within each country
  • EDUCYRS: Education I: Years of schooling
  • DEGREE: Education II: Highest education level
  • INSCHL: Education I sub: Still in school
  • EMPREL: Employment relationship
  • FULLPART: Working full- or part-time
  • WRKHRS: Working hours per week
  • EMPCNTRT: Employment contract of wage employees
  • typorg1: Type of organization/firm, profit-non-profit
  • typorg2: Type of organization/firm, public, private
  • cn/jp/kr/tw_rinc: Earnings from jobs (either both main and part-time jobs, or just main job): per country
  • cn/jp/kr/tw_rincx: Income other than (main)jobs: per country
  • hhdnum: Number of household members
  • cn/jp/kr/tw_hinc: Total household income: per country
  • rltvhinc: Household income compared
  • hmalone: Household member: Living alone
  • ttlcnum: Total number of child


Groups of variables

The remaining variables are largely dependent variables – though some may be used as independent – surveying a multitude of topics listed here. The bolded groups were the ones I chose to explore deeper into by (1) doing basic exploratory analysis and (2) analyzing it in relation to some of the demographic charactistics listed above:

  • TrustV71:V88: these are trust level regarding different subjects (ranging from friends/relatives to colleagues/community leaders/strangers). Individual view on trust/distrust in people is also measured. Variables like V89 (“years living in the same place”) may be used to analyze these.
  • Socializing behavior/occasions — V33:V37: items in this group capture the nuances of eating out or social drinking/eating (how often, which status speaks first, are seatings carefully managed, etc)
  • Financial ownership — V91, s5_1_cjt:s5_3_jt: items here measure ownership of different financial instruments. These items may be used to analyze variables in contact/network and status perception categories.
  • Group Participation — V4:V15: items include information ranging from participation in different kinds of groups (political association, residential/neighborhood association, social service club, consumers’ cooperative group, religious group, alumni association, recreational assocation, labor union, trade association, etc) to the specific orgnization the respondent participated most actively in the last year, as well as dynamic within the organization (homogeneity and “hierarchical relationships among members”). What’s to be noted here is that some of the variables may serve as independent variables in analyzing variables in other groups (i.e. in seeing how the number of foreign contact influences certain behaviors/trends)
  • Status perceptionV45:V48: these gauge the so-called “social tolerance” of the respondents regarding those who have lower, equal, or higher social status.
  • Contact/networkV16:V32, V38:V42: these items range from the number of people one contacts in daily life, contacts living abroad, job-search related network, occupations of regularu contacts (professor, lawyer, programmer, hair dresser, policeman, etc), to people to ask for help when different issues arise (psychological/health, financial, emergency/natural disaster).
  • Neighborhood dynamicV43:V44, V69:V70: these variables rate neighborhood environment as well as number of neighbors that the respondents are able to contact and/or greet.
  • Political advocacy — V57:V68: these items gauge experience of political outreach and the respondents’ political interest and views on citizen influence and complication in politics.
  • Community engagement — V48:V55: frequency of participation in different types of community meeting and engagement in volunteer activity of different causes are covered in these variables. Variables like V90 (“power to make important decisions to change life”) can be analyzed in conjunction with these variables.
  • Natural disaster security — s1_jptw:s4_jptw, V56: variables here indicate (for Japan and Taiwan only) the perceived possibility of earthquake, flood, landslide/avalanche, accidents, as well as community cooperation at natural disaster. These items may serve as interesting independent variables in analyzing trends in trust, financial ownership, and community engagement.
  • Other key insights: happy and health that report self-rated measures of happiness/health with potential to provide key insights in regards to how these social dynamic factors play out in individual lives.



Count of entries per country

Our analyses will contain exploration of trends and patterns among the four countries as well as within each country. Our dataset contains the total of 11,684 observations. Here’s the breakdown:



Now, one thing to consider here is the fact that these figures are not proportional to the size of the population of the four countries. In the year the survey was conducted, China had 1.351 billion1, Japan 127.6 million2, Korea had 50.2 million3, and Tawian had 23.7 million people.4


Though there may be difficulties in interpretation / bias that may arise due to the discrepancy between the relative size of true population and the ratio among sizes of survey sample, for our purpose of the study, we are going to skip the weight configuration among the countries and only account for the within-country sampling error/nonresponse bias as I will discuss below.

EDA of demographic variables

The following is a series of brief overviews of some of the interesting demographic indicators. I begin with a basic exploration examined by country, and deep diver into how some of the demographic factors interact with each other. The key thing to note about the deeper analysis in this section is the fact that the interaction analysis of these demographic factors may not be consistent information available on the census-level. Though the WEIGHT variable exists to account for few of the discrepancies, it should be known that the weighting may play just a limited role in making the analysis true to census level data. Thus, though there may be extrapolations and analysis done in conjunction with the countries’ demographic characteristics, this section ultimately explores the demographics of the survey respondents with slight adjustments to their age group and sex.

Age

All four countries



Shown above is the apread of age of respondents of all four countries. You may be familiar with the typical age spread of developed countries from middle school geography courses, most commonly seen in population pyramids, which PopulationPyramid.net provides interactive graphics for. The discrepency between our respondent age distribution and that of the census level data is quite evdient, especially for people in their 30s-40s. To account for this, I am going to use the WEIGHT variable provided in the dataset that adjusts for nonresponse bias and sampling error.



It is difficult to tell how the WEIGHT variable improved the spread, reasonably since the four countries are on significantly different places in the demographic transition, which brings us to the point that I wanted to emphasize through this EDA: how different the countries of East Asia which are often grouped together in social and behavioral scienices. Let’s take a look of the age distribution at the country level:

Per country



Contrasting the weighted graph to the unweighted, we can tell that the WEIGHT variable increases the portion of younger respondents (in 20s and 30s) to match each country’s census data. The codebook notes that the weights also account for potential error/bias in gender as well for all four countries, and urbanization and education levels as well for Taiwan.

Korea and Taiwan’s distribution of age are relatively skewed to the right; we can see that the spread is consistent with age distribution/composition of developed countries. What is notable in Japan’s graph is the the fact that older population (people in their 60~80s) constitute a significant portion of the respondents. This is reflective of the high lifespan of Japan at 83 years in 2012, compared to the other countries (81 for Korea, 75.6 for China, 79 for Taiwan).5) This also reveals as well as the aging of the society as the country was experiencing at a higher rate compared to the other countries due to the low death rate and the incredibly low fertility rate, though Taiwan and Korea have even lower fertility rates.6 In the case of China, the drastic peak in the 40s group is consistent with the rapid increase of birth rate 1960s (during the post-war period) that constantly decreased until dipping (and increasing slightly back again) down due to the implementation of the one-child policy. This sample still fails to reflect the second peak in the people in their 20s, presumably one generation down from those born in the 60s.7
It should also be noted that the median age differs somewhat significantly country by country:

Country Median Age
(1) CN-China 48
(2) JP-Japan 54
(3) KR-South Korea 50
(4) TW-Taiwan 45




This plot illustrates the point made earlier about Japan’s aging society of the points mentioned above; Japan’s median age is among the highest in the world (along with Germany) at 46.1.8

Status

Let’s take a look at other demographic factors such as marital status, main status, and education (keep in mind that this survey only studied adult citizens of the countries):

Marital status



Given the fact that surveys were conducted via different institutions in each of the countries, it is difficult to tell whether or not the discrepancy among these proportions come from the fact that they are from different countries with widely different cultural/social norms, or if they come from the difference in sampling methods. Nevertheless, the majority of repondents from all four countries were married. Second biggest share of marital status was those who were married, though not as significantly big in China. China, at 79%, had the biggest proportion of respondents who were married. Taiwan had the lowest, at 58%.

Though it may not eliminate the potential inconsistency from sampling discrepancy, more valuable information might be discovered by comparing these variables with one another, or with other demographic factors like age and gender.

Marital status breakdown by sex



The gender breakdown of marital status reveals some interesting patterns. Some of the most conspicuous findings is that more men have never been married, and that the widowed population is dominated by females, most of whom were in their 60s and 70s. The biggest age group among widowed men was slightly older, in their 80s, consistent with the fact that females are widowed earlier due to relatively longer age expectancy. Though the four countries show farily consistent gender shares of different marital statuses, striking deviation can be seen from Taiwan’s female dominance (n = 8 female vs. 2 male) among the those that are separated with the intention to divorce (what do the men that they used to be together with these women perceive their marital status as?). Japan’s cohabitance group being entirely male is also interesting, it was just three respondents in their, an interesting figure to ponder upon given the fact that two parties must be separated The fact that more women with from the norm when looking at

Marital status breakdown by age group



There are very few people in their 20s married in Japan and Taiwan when compared to China and Korea. Japan is also an anomaly in showing a robustly high count of married people among the older population throughout mid-to-late years as well, while the remainder of the countries see a steep decrease after hitting a peak in 30s and 40s.

Never married



Here, we are taking a look specifically at those who were never married. There’s obviously a high count for the younger demographics, with the exception of Japan, where there’s a lot more never married people in their 30s. In Japan, most never married people are male until it reaches late 70s and on, when there are presumably less males counterparts existing. This similar trend is also seen in Taiwan. What’s interesting about China is that though there is a lot more unmarried younger females, older population that were never married are all male; this a similarly cased in Korea.

Main status



More than half of valid responses came from people in paid work for all four countries. Second most common status was being in domestic work or retired for every country except for Korea, where domestic work and “other” were most common after paid work. Let’s take a look at how this breaks down by sex:

Main status breakdown by sex



Indisputable findings are the dominance of femaes in domestic work, and more than half of those in paid work being male. The retired population was also dominated by male with the exception of China, where there were slightly more females indicating their status to be retired. Female’s main status China actually stands out in other regards as well, in the sense that fields dominated by male in other countries (like retired, apprentice or trainee, unemployed and looking for a job, etc) are fairly even in the case of China. A noticeable portion of male are also in domestic work, and there are more women in education than men.

Domestic work



Taking a closer look at those whose main status is domestic work, we see different patterns arising in different countries. With the predominant pattern being that there is most females in their mid-ages (centering around the 50s age group) in domestic work, Japan is an evident exception here, as it showcases a bimodal distribution with peaks in the 40s (which can be overlooked, for China and Korea has high counts for women in their 40s here too) and 60s-70s (seen in no other countries). China’s highest count being women in their 30s is also notable.

Education level



The most predominant education level is one of the most evident discoveries from this plot. Korea had the highest share of university graduates and overall post-secondary education attainment. In the case of Taiwan and Korea, the two highest education level are high school degree and university degree, the two showcasing an even share of counts. China had the lowest rate of those who completed what is considered “compulsory education” (attaining high school diploma), with the majority of respondents having education level in between no formal education at all and completing junior high, which had the highest share. Japan had a remarkably high proportion of respondents who obtained until high school degree, and overall had the highest share of respondents who completed compulsory education and beyond. It is a known fact that vocation and professional pursuits are prioritzed/respected before an academic degree in Japan, as also indicated it having the highest share of those who attained junior college degrees.

Education breakdown by sex



Japan had more noticeable differences from the other countries in the breakdown by sex as well, as junior college degrees were received predominantly by females (fairly even in the other three countries), while university and graduate school degrees were largely clainmed by men (university degree and even graduate school, with the exception of Taiwan, also had quite even ratios in the other countries). The group that did not receive any formal education (0) was dominated by females, and the majority of those who only attended up until elementary school (1) was also female for all four countries.

Perceived social hierarchy

Perceived social hierarchy by country



The four countries have surprisingly similar perceived social hierarchy of the self (this may be one of the first variables whose pattern coincided for all four countries!). The only noticeable difference is in the fact that Japan‘s median score was 6, an interesting statistic considering that all of the other countries had a median score of 5. Taiwan also showed interesting findings, such as a relatively high portion of respondents from Taiwan rating their social hierarchy as 1 and 7 (the other three countries’ second common score was one lower than the median, not higher).

Perceived social hierarchy by age group



While most countries experience the “worsening” of perceived social hierarchy as you observe bottom to top across the y axis (younger to older), Japan showcased the opposite trend, with greater share of people rating their social hierarchy to be 6 or higher as you went up the y-axis.

Perceived social hierarchy by sex



Taking a look at the gender breakdown, with the exception of Korea, where females perceived their social hierarchy to be the lowest as well as highest (with males clustering in the middle ranges), the countries saw a diagonal divide between the genders: more male perceived their social hierarchy to be low (close to 1), and more females perceived their social hierarchy to be high (close to 10). Minor exception lies in Taiwan’s almost even gender divide between those tho rated their social hierarchy to be 10.



Overall, though, the most noticeable trend is that females rated their social hierarchy in the middle (5 or 6) at a share significantly higher than males. In the case of China and Taiwan, significantly more males than females rated their score to be fairly low (4 or lower), though most males in Japan and Korea rated somewhere around the middle ranges (4 to 7).

Perceived social hierarchy by education level



Interesting findings can be found looking at perceived social hierarchy broken down by education level as well. One of the most striking findings is that though university + graduate school degree did increase higher as you went up the y axis (lower to higher perceived hierarchy), this trend generally ended at around the score of 7 to 8. The highest social hierarchyof the self (of score 10) was perceived, surprisingly, mostly by those with high school diplomas (middle school in the case of China and Taiwan). It is interesting to conclude that education level did not ensure the highest perceived social hierarchy.



In terms of community level, Taiwan showed city residence (as well as suburban residence) was positively associated with increasing perceived social hierarchy. In no other countries were there distinct pattern as such caused by community type.

Career



Excluding outliers (filtered to be less than the maximum number of hours per week), the median weekly work hours for China was 47 hours, 40 for Japan, 51 for Korea, and 44 for Taiwan. Mean work hours is similarly high with China at 52 hours, Japan at 41 hours, Korea at 52 hours, and Taiwan at 47.

The breakdown of the numnber of people working full time vs. part time jobs is as follows:



China has a notably small proportion of workers working part-time jobs, while Japan has a big proportion. Looking at the number of hours worked by full time vs part time workers reveals more interesting discoveries:



These figures generally get higher as we filter to those working in full-time jobs only. However, interestingly enough, Taiwan and China’s median work hour actually decreases here from 47 to 46 hours, indicating that there may be long hours worked for those working part time jobs. On the reverse side, the significant increase in median work hours for Japan indicates the short hours worked for those working part time jobs. Among the respondents, The median for Korea here is starkly high at nearly 54 hours; to specifically elaborate on Korea, where my parents worked for several decades, extension of a regular workday to one with a night shift (with no overtime pay in most cases) is a regular occurrence.

To see if our assumption about part time work is true, let’s take a look at those working part time jobs:



China has remarkably high working hours for part-time workers, with the median hours at 47 and the mean hours at 51.65, less than an hour less than the average work hours for full-time workers. One assumption that can be made about this is that a significant portion of part-time workers may be working more than one job. It might also be the case that the notion of full time and part time job may be different in China It should be noted here that our dataset does not capture whether or not a respondent was working a part-time job in addition to his or her full-time job, which could’ve led to a more interesting analysis on this matter.

Work hours vs. gender


Generally speaking, females work longer hours than male. This exmaple is most illustrative in the case of Japan, where the lower hours are predominantly blue and the longer hours predominantly red.

This is especially interesting when you note the fact that almost half of Japanese females are working part-time jobs:

Breakdown of Japanese work type by sex

(1) Male (2) Female
(1) Working full-time 661 331
(2) Working part-time 119 318


In no other countries do females have such a high percentage of part-time workers, as can be seen in this plot:




Looking at the breakdown of specifically Japanese female workers, we can see that what was driving the gender discrepancy in Japan’s “Work hours by gender” graph above indeed was female Japanese part time workers.

Gender wage gap



The trend is conspicuous for all four countries: females work less hours, and earn less money. From the cluster on the lower left corner, we can also tell that females get paid less for the same number of hours worked; vertically scanning the plot from top to bottom (ignoring NAs), there are very instances where a blue dot precedes red dots, nor is there a range of x values where blue dominates the top portion. (This may be our second finding in which there was a consistent pattern across all four countries).




EDA of social capital indicators

Group participation


Though participation in none of the listed organization groups was dominant in all four countries, there are interesting differences among the four countries. First of all, 4442 out of 5766 valid responses for China indicated that they did not actively participate in any of the listed organization types. Korea did not have anyone indicating “None of them,” but assuming the lack of answer as nonparticipation just specifically for this question, the highest “active participation rate” in any of these organization was Korea (75%), followed by Japan (63%), Taiwan (31%), and China (23%). To my surprise, China had the highest active participation in political associations (N = 342), a fact highlighted by the fact that political associations had the lowest participation (nearly in the case of Japan, where citizens’ movement/consumers’ cooperative group was lower by 11 respondents) in the three other, democratic countries. Recreational association (07), alumni association (06), and religious group (05) saw relatively high active participation in most of these other countries, but Korea had starkly high counts for all three, which speaks to how the high “social-ness” of Korea plays out in the Korean society. In the case of alumni association, regular meet-ups with college, middle and high school, as well as elementary school friend-groups and official alumni groups is quite common and regular in Korea. The relatively high count of residential/neighborhood association (02) may speak to the circumstances of residing in apartment complexes, where, in the case of Korea at least, residents are semi-required to attend regular meetings and encouraged to volunteer to serve in leadership roles or in “women’s associations.”

This leads us to the breakdown of organization participation by gender:

Group participation by gender



The active participation rate does not differ as drastically between the two genders as been expected, though there are a few notable findings. As discussed above, participation in residential/neighborhood associations is dominated by women in Korea (67% female), whereas it is fairly even between the genders in the other three countries. Political association (1) is dominated by male in all four countries, and labor union (8) is expectedly dominated by male in all countries, albeit with a noticeabley high percentage of females in the case of Taiwan (44%).

Group participation by age group

Let’s take a look at active participation rate by age group:



Despite the fact that groups in 20s are few in number (as seen in the exploratory data analysis on the age distribution above), respondents in their 20s dominated the scene in recreational (hobby and sports) associations (7). Residential communities (2) are most actively participated by the older participation with the exception of Japan, where participation is almost evenly distributed among age groups. Looking at the difference among countries in terms of where people in their 30s-60s are most actively participating in reveal insights into the workforce that are related to the difference in retirement age we discussed above. Though dominated by people in their 20s in the other countries, recreational associations (7) are most actively participated in Japan by people in their 60s, followed by those in 70s and 50s. Labor unions (8) are most actively participated by people in their 30s in Japan, an interesting contrast from Taiwan and China where participation is highest among older workers in their 40s to 50s.

Group participation affected by hierarchy / homogeneity within group

How does perceived hierarchy affect an individual’s participation in an organization? What about in the case of homogeneity – are individuals more likely to participate if they find groups homogenous? To answer this, sum up the participation scores in the nine kinds of organization/association, recode them so higher score means more participation, and compare this figure to their scores of perceived hierarchy and homogeneity.

It should first be noted here that, for both homogeneity and hierarchy, we do not know the scores for each kind of organization. We only know the general figures for all organizations, worded in questionnaires as “[hierarchical relationships/homogeneity] among members in organizations respondent participated.” Additionally, we created the “participation score” by summing up all the individual participation scores of different types of organizations, erasing the nuances of who participates more in what kind of organization(s). Though there is implied eradication of the differing levels of homogeneity/hierarchy among the different organizations respondents participate in and also among the participation levels of different types of organization, we can make an informed assumption about the characteristics of those who, on average, do participate more in more kinds of organizations.

Hierarchy



Generally, people who saw hierarchy to be less present in their organizations also participated less (the graphs are heavy towards bottom left), perhaps indicating a natural aversion to organization-dynamic and the somewhat inherent existence of hierarchy in organizations. The average participation score for each hierarchy score supported this sentiment, as those who saw more hierarchical relationships were likely to be participating generally more in organizations (seeing the general increase in average participation score (marked in red) as hierarchy score goes up). Another way to interpret this trend could be that the more organizations people were involved in, the more likely it was that they were exposed to the hierarchical dynamics in organizations.



Taking a look at the cumulative level graph for all countries, the second interpretation may make more logical sense, as the highest participation scores are seen in lower hierarchy scores (as in, people don’t necessarily join more organizations because they like hierarchy).



Homogeneity



Perceived homogeneity is fairly high (biggest dots in homogeneity scores 3 and 4) for all four countries, but this does not necessarily mean that more homogeneity mean more active participation in groups, as can be seen in the fact that there is no particular trend between average participation scores going across the homogeniety scale. Alikeness ensures more active participation just slightly in the cases of China and Japan, but not so much in the other two (there actually is a noticeable drop of participation score in the case of Taiwan).



The lack of trend between homogeneity score and participation score in all countries supports the claims above, perhaps debunking the common perception that east Asians, that are known to be ethnicially homogeneous, prize homogeneity as well as community ideals exercised within homogeneous groups.

Contact/Network

Family & relatives



Looking at the spread of the number of family members/relative individuals interact with on an ordinary day, a similar spot can be named across the four countries: the number of contact peaks at 1-2, and decreases step-wise. Proportion of those who answered they interacted with 0 family members/relative on an ordinary day is significant for Japan (at 30%) and Taiwan (31%) though, especially compared to Korea (at 14%).



Looking at the measure broken up by gender, we see a clear dominance of female in having more daily interactions with family members. Males are more likely to not have any interaction with family member/relative on a given day.




Looking at the age breakdown, we can see that the highest average family contact (>1) occurs in one’s mid ages around 30s-50s (significantly higher in the 30s in the case of Korea), slightly higher for females than it is for males. From one’s 40s, the average family member contact gradually decreases / lack of contact gradually increases. In one’s 90’s though, the contact with one to two family member suddenly increases.

Outside of family/relatives



The four countries show somewhat drastically different trends for the number of contacts with people outside the family made on an ordinary day. In the case of China and Korea, the trend from family/relative of peaking at 1-2 people and gradually decreasing remains. For Japan, there is a “peak” at 1-2 people, but the spread is quite uniform until the 50-99 people mark. Taiwan is the most astonishing case of the greatest number of people having around 10 to 19 ocntacts, and the spread looking almost normal around that presumable “mean.”


The gender breakdown shows a contrasting case from case of contacting family member/relative. While females had more contact (in almost every category of contact ranges), here, more males have greater number of contacts (from 10 contacts and up). This pattern is especially strong for Japan and Taiwan.



A clearer, more negatively linear relationship can be seen when looking at the age breakdown. In all four countries, the younger you are, the more contacts you are likely to have, and the older you get, the more likely it is for you not to have contact with people outside of your family on an ordinary day. This makes adequate sense given the social environments we are provided with while in school, while working (or staying at home raising children), and while aging, that lead towards general social isolation.

Number of neighbors in greeting terms

By country



Japan has a spread generally uniform across all ranges, with the highest range being 3-4 neighborhoods. The rest of the three countries reported most commonly having 10 or more neighbors respondents were in greeting terms with.

By gender



For Japan and Korea, females had higher number of neighbors they were in greeting terms with, and males had generally less neighbors they were in greeting terms with. However, the reverse was the case for China, where a significantly higher number of males had 10 ore more neighbors they were in greeting terms with. In the case of Taiwan, the gender split was fairly even all throughout, though males leaned towards having more neighbors they greeted compared to females.

By age group



Across all four countries, there was an evident pattern with age and the number of neighbors respondents greeted with. Generally, the older you were (until around 80s-90s), the more greeting-terms-neighbors you had. The peak is lower in the case of China, where people in their 60s had the highest number of greeting-terms-neighbors.

Social dynamic with nonkin contact

A closer look at the nonkin contacts can be attained by looking at the social status of the contact, relative to the respondent, as preceived by the respondent.


All four countries show that social contact is most predominant with those who people perceive to be socially equals.

Country Proportion of (2)
(1) CN-China 0.8922667
(2) JP-Japan 0.8130759
(3) KR-South Korea 0.8381089
(4) TW-Taiwan 0.7937868

Among those, China had the highest share (89%) of those who answered that the majority of people they contact with are socially equivalent to them.

In all four cases, there are more contacts with those who people perceive to be socially higher than lower, with China having the biggest margin in between the two kinds of contact. The ratio of those who have more contact with people with higher social status compared “(3)” to those with those who have more contact with lower social status “(1)” is as follows:

Country (3):(1) Ratio
(1) CN-China 5.965517
(2) JP-Japan 2.520661
(3) KR-South Korea 1.790123
(4) TW-Taiwan 1.916667


China stands out in this figure too, as about six times the people said they have more contact with people higher social status than lower. The ratio was significantly lower for the other three countries.



Looking at the proportion divided by age group reveals interesting findings. Though it is unclear whether or not these are generational differences or the social circumstances depending on age that affect thre relationship between age and social dynamic with your contact, it can be seen that generally speaking, contact with people who are “socially higher” than you decreases the older you get, with the exception of China, in which the count of socially higher contact remains almost constant throughout the age groups. Noticeable difference exists in the share of socially lower contact though; in the case of Korea and China, social contact with those with lower social status dramatically increases in the 90s age group, presumably due to the influence of time spent at senior citizens’ center (an institution that is comparable to daytime retirement home, usually run by the government (free of cost)). This contrasts with the cases of Japan and Taiwan, where this share is the highest for those in their 40s and 50s.

Trust and help

Trust in people

By country



The four countries seem to have generally similar patterns of trust in people, though Korea and Taiwan lean to be more skeptical (mode: 3, “You usually cannot be too careful in dealing with people”) than China and Japan (mode: 2: “People can usually be trusted”).

By gender



More females were relatively skeptical of people across the board. More males than females from China, Korea, and Taiwan were likely to be trustful of people.

By age group



Several interesting patterns can be seen while looking at how different age group responded to how trustable people are. China and Korea show a positive pattern between age group and overall trust level, indicating increasing leniency, while Japan and Taiwan show that more older people show distrust of people as you go up the y-axis. Korea is an interesting case as a stand alone too; as you go up the age group, more people indicate that people usually can be trusted (2), but also increasing is the reponse that you can almost never trust people (4). Younger Koreans’ predominant response that you usually cannot be too careful in dealing with people (3) is also interesting, as all other countries’ teens (and even 20s) generally responded with higher trust of people.

Trust in self agency

This variable measured respondents’ belief in their own power to make important decisions to change the course of their lives.

By country



All four countries predominantly had a wishy-washy mode of 3, “somewhat able to change life”. Taiwan was skewed significantly positively, with almost double the number of people answering mostly able to change life (4) compared to somewhat unable to change life (2), which was the second popular option for all three of the remaining countries.

By gender



Though relatively subtle, a really interesting finding is that across the board, significantly more males had the belief that they were mostly able to change life (4), while more females answered pessimistically that they somewhat unable to (2) and mostly unable to change life (1).

By age group



Age group also showed a consistent pattern across all four countries. Going up the y-axis (younger to older), positive answers ((4), (3)) declined and negative answers ((1), (2)) increased. Though younger people showed predominantly positive belief in self-agency to change one’s own life, this belief dwindled across the age groups, and eventually got dominated by more skeptical beliefs. Japan and China, however, had significantly less proportion of negative beliefs among the older population.

Financial vs. psychological help

By country



All four countries show somewhat similar patterns of where to go for help for financial and psychological difficulties. Family took up a great portion for both financial and emotional/psychological in all four of the countries. For financial help, other relatives and friends (though much lower than family) were popular second chocies. For psychological help, excluding those who didn’t seek help, friends was the second popular choice for all four countries, but Korea and Taiwan showed that they turned to friends almost as much as they turned to family members. An important fact to note here is that neighbors (4) and professional workers/institution (6) and were significantly the least popular choices in all four countries.

By gender



In the case of financial help, gender revealed a consistent and somewhat significant pattern for all four countries. A lot more females were likely to seek help from co-residence family and other relatives, whereas in the case of males, many of them turned outside their family to friends and work colleagues. More males also likely to indicate that they never had such a problem (9).

Emotional help had a somewhat of a different pattern between the genders. A clearly consistent pattern across the board was that way more males were likely to claim that they never had such a problem (9) or simply did not seek help from anyone (7). Way more females turned to other relatives for help in all four countries, while those turning to co-residence family was split between Taiwan-China (more males turning to family) and Japan-Korea (more females turning to family, though not by much) pairs.

By age group



In the case of emotional help, the family-friend dichonomy is extremely visually striking. An overwhelming portion of younger respondents reported turning to friends for psychological help. This portion gradually drops as you go up the y-axis, as the portion relying on co-residence family member gradually increases, until it reaches the most popular choice among older groups. There’s also a greater proportion of people not seeking for help from anyone (7) as you go up the age group with the exception of Japan, where older groups were instead more likely to say that they never had such a problem (9).



Conclusion

Through this exploratory data analysis, many differences among the four East Asian coutries so often grouped together in the Western academia were explored. First, we went through the demographic indicators such as age, sex, marital status, occupational status, education level, and etc to explore how the respondents from the four countries differed from one another. Among the key difference was education level (highest degree obtained), in which we found the high volume of people with college degrees in Korea, the predominance of middle school graduates in China, and the extremely high completion rate of compulsory education (obtaining high school diploma) in Japan. Crossing education level with one’s perceived social status, we also discovered that post-secondary education limitedly guarantees perceived social hierarchy to mid-to high range. Looking at factors related to career and work hours, the first similarity we found among the four countries is how gender seemingly determines one’s wage when controlled for work hours.

We then moved onto analyze the social capital variables the study was designed to survey. Among the many groups of variables, we specifically looked into group participation, network/contact frequency, and trust in self and others. In this section, more prevalent than a country-level similarity were trends found by sex and age group across the four countries. For instance, the increasing skepticism regarding the general trustworthiness of people as the age of the respondent increased was a pattern found across all countries. Likewise, factors like trust in self and others, who people turned to for financial and/or emotional help were largely guided by the gender one identified with and/or the age group the respondent was in, rather than on a country-by-country basis.

What attests to the overall lack of generalizable characteristics among these four contries was that there were no prevalent pairings of countries across the analyses, despite the fact that there is presumable historical and political background for China and Taiwan, as well as for Korea and Japan that may have influenced the countries’ social capital varibales to this day. I hope that my analysis of the demographic indicators and social capital variables carried the message across that, despite the common notions surrouding these “East Asian countries” in social and behavioral sciences in the Western academic world, it is practically difficult to spot predominant patterns that guide the four countries in one way or another. There exist differences, and nuanced differences that this dataset, even upon more robust analyses of more skilled data scientists, are unlikely to be able to cluster together.


  1. World Bank Open Data: Population, total

  2. World Bank Open Data: Population, total

  3. IMF Datamapper

  4. World Population Review

  5. [World Bank Open Data: Life expectancy at birth] (*https://data.worldbank.org/indicator/SP.DYN.LE00.IN*)

  6. “The Aging Population and Aged Society,” ILC Japan, 2013, pp. 10

  7. Karklis, Laris, and Richard Johnson. “How China’s Population Has Changed since 1950.” The Washington Post, WP Company, 29 Oct. 2015

  8. Khosla, Simran. “These Maps Show Where the World’s Youngest and Oldest People Live”. Public Radio International, 8 Sept. 2014